Open cbizon opened 1 year ago
Thanks @cbizon. The relevant KP source is this new one we added recently:
https://biothings.ncats.io/rare_source
and this relevant issue comment regarding GARD IDs used in this KP API
We will explore a bit to see where we can have a good mapping file for GARD IDs.
cc'ing @erikyao @colleenXu
It looks like there are lots of mappings in MONDO, so that might be the simplest thing
cc @sierra-moxon in case there's something we should know about the MONDO mappings
:+1: That will be easiest if MONDO already has the mapping.
Note that GARD has created a new rare disease list they recently supplied to the Mondo team. We have already integrated it into the ETL, so you should see a lot of updated GARD mappings in the next Mondo release cc @nicolevasilevsky
Another related todo item is to add GARD as a supported identifier under BioLink Disease class.
we noticed that DOID (disease ontology) seems to have some GARD xrefs? https://disease-ontology.org/?id=DOID:3138
Note that GARD is undergoing major changes right now - they are compiling a completely new list, which means that many of the old IDs (I counted at least 400, but I may be wrong) are now deprecated.
GARD IDs are derived from ORDO, or in some cases, OMIM. So you can assume that when there is a GARD-ORDO mapping or a GARD-OMIM mapping that is exact as per our SSSOM conversion (https://github.com/monarch-initiative/gard/releases/tag/v2023-06-08, see gard.sssom.tsv), the mapping is 100% correct. That is the safest to use.. For Mondo, in the next release, you will see all the GARD IDs mapped through their ORDO and OMIM associations so when a Mondo ID has an exact ORDO mapping and a GARD ID has that same exact ORDO mapping, the Mondo ID is declared an exact match to the GARD ID).
Because of this, I don't think there is any need to trust a manually curated GARD mapping from any source, since they derived the terms!
GARD is a rare disease vocabulary, and @newgene needs it in nodenorm to ingest a particular data set.